Multidimensional Analysis of Geographic Genetic Structure

Publisher Website
Google Scholar
Abstract
Multilocus data are an important source of information on genetic variation within and among natural populations. Hierarchical techniques are used widely for examining multilocus data. However, the limitations of hierarchical (phenetic or phylogenetic) algorithms to depict geographic genetic structure, particularly within species, have long been recognized. Multidimensional scaling of genetic distances is herein examined as a useful technique for exploratory analysis of geographic genetic structure. The advantages and limitations of multidimensional scaling are discussed and illustrated with reanalyses of two case studies: pocket gophers (Thomomys bottae) of the central California “genetic group” and a hybrid zone of two chromosomal forms of the tent-making bat (Uroderma bilobatum). Multidimensional scaling does recover hierarchical patterns when present and is especially useful to uncover nonhierarchical patterns of variation. The finding that reticular and clinal patterns of variation may be examined via multidimensional scaling even in the absence of fixed genetic differences between hybridizing taxa opens new possibilities for studying geographic genetic divergence and speciation.